Human body posture estimation method and system based on attention multi-resolution network

A multi-resolution, human body posture technology, applied in the field of image processing, can solve the problems of low accuracy of human body posture recognition, and achieve the effect of accurate human body posture estimation results and high spatial positioning accuracy

Pending Publication Date: 2021-06-22
SHANDONG UNIV
View PDF9 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Since different resolution features contain joint point information of different scales, the existing research results often focus on how to better extract multi-resolution representations, while for the fusion of multi-resolution representations, the method of adding corresponding position elements is directly used, which leads to different The difference in the importance of resolution branch channel information is ignored, which makes the accuracy of human pose recognition lower

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Human body posture estimation method and system based on attention multi-resolution network
  • Human body posture estimation method and system based on attention multi-resolution network
  • Human body posture estimation method and system based on attention multi-resolution network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] As a task with high spatial sensitivity, human pose estimation is of great significance to improve the spatial positioning accuracy of feature information at different resolutions. Shallow high-resolution features retain more local and detailed information, which can be more accurate. Good for capturing small-scale human bodies; deep low-resolution features contain global information and classification capabilities, and are more suitable for capturing large-scale human bodies. How to extract and fuse effective features contained in different resolutions is still an open problem in the task of human pose estimation. The current method focuses more on the feature extraction method of the network. The fusion stage is often just a simple addition of corresponding position elements. This has the problem of unreasonable fusion of multi-resolution representation information. In order to solve this technical problem, improve the human body pose estimation method. Accuracy, in t...

Embodiment 2

[0070] In this embodiment, a human body pose estimation system based on attention multi-resolution network is disclosed, including:

[0071] An image acquisition module, configured to acquire a target image to be identified;

[0072] The attitude estimation module is used to input the target image to be recognized into the trained attention multi-resolution network model to obtain the attitude estimation result;

[0073] Among them, the attention multi-resolution network model includes a fast sampling stage, a main part of the network and a representation fusion module. The fast sampling stage down-samples the input image and extracts representations of different resolutions, and extracts resolutions from representations of different resolutions through the main part of the network. Rate branch features, the representation fusion module uses the channel attention mechanism to weight and fuse branch features of different resolutions to obtain fusion features, and perform pose e...

Embodiment 3

[0075] In this embodiment, an electronic device is disclosed, including a memory, a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, a method disclosed in Embodiment 1 is completed. The steps described in a human body pose estimation method based on attention multi-resolution network.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a human body posture estimation method and system based on an attention multi-resolution network. The method comprises the steps: acquiring a to-be-recognized target image; inputting a to-be-recognized target image into the trained attention multi-resolution network model to obtain a posture estimation result, wherein the attention multi-resolution network model comprises a rapid sampling stage, a network main body part and a representation fusion module; in the rapid sampling stage, performing down-sampling on an input image, extracting representation information of different resolutions, extracting resolution branch features from the representation information of the different resolutions through the network main body part, so as to represent that the fusion module uses a channel attention mechanism to carry out weighted fusion on the branch features with different resolutions to obtain fusion features; and performing attitude estimation through the fusion features. And accurate estimation of the human body posture is realized.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a human body pose estimation method and system based on an attentional multi-resolution network. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art. [0003] The visual system is the most important way for humans to observe and understand the world. Statistics show that the information obtained by humans through vision accounts for at least 80% of the total amount of information. For a long time, researchers have hoped to simulate the human visual system by computer, so that the machine can accurately identify and locate the target object in the image, and finally realize the understanding of the advanced semantic information hidden in the image. Analyzing human behavior through computer vision can greatly improve the convenience of human...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V40/20G06V10/464G06N3/045G06F18/253
Inventor 常发亮丁锐李南君蒋沁宇
Owner SHANDONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap